In many current energy measurement devices (e.g., metering devices), an acquisition buffer of a fixed amount of time is used for sampling voltage and current signals. This fixed duration of time typically accommodates a fractional (non-integer) number of line cycles (e.g., half or full line cycles), which causes, for example, measurements of metrological values from the acquisition buffer (e.g., root mean square (RMS) determinations, energy measurements, etc.) to fluctuate. Previous and current solutions attempt to remedy this problem by averaging several successive time periods worth of data, making measurements on a partial buffer, and/or applying windowing to the associated Fourier transform. However, these solutions can be slow and expensive due to the additional computation(s) and the additional memory that may be required. Further, these solutions may not provide consistent results.
There are other issues that can occur when sampling voltage and current signals for energy measurement. For example, to be cost effective, it would be beneficial for energy measurement devices to use less expensive sigma-delta (or delta-sigma) analog-to-digital converters (ADCs). In general, a sigma-delta ADC encodes a received analog signal using high-frequency sigma-delta modulation, applies a digital filter, and outputs a higher-resolution but lower sample-frequency digital signal. Sigma-delta ADCs are intended to operate at a fixed frequency, however. If a sigma-delta ADC incurs a frequency rate change when the ADC is active, the output samples from the ADC may have significant errors due to the effects of the particular internal modulators and filters used in the ADC. For this reason, sigma-delta ADCs are not typically used in designs where frequency changes, or phase changes, will occur or are required/expected. Typically, a Successive Approximation Register (SAR) ADC would be used instead.
Solutions that overcome these signal sampling issues are needed, in particular for energy measurement devices and other types of devices that require highly accurate output.
In the drawings, the leftmost digit(s) of a reference number may identify the drawing in which the reference number first appears.